CA3071547C - Method for assisting at least one movement of a user and corresponding device - Google Patents

Abstract

The invention relates to a method for assisting at least one movement of a user by means of a robotic device (1), the method allowing the robotic device to be operated in the following two modes: - Default mode: controlling the force of the robotic device via a computer (4) of the robotic device; - Measurement mode: controlling the path of the robotic device, via the computer, and saving in the computer at least one force exerted on the robotic device by an environment that is external to said robotic device in order to implement the force control of the default mode.

Description

ASSISTANCE PROCESS FOR AT LEAST ONE MOVEMENT OF ONE
USER AND CORRESPONDING DEVICE
The invention relates to a method of assisting minus a movement of a user. The invention relates to also a corresponding device.
TECHNOLOGICAL BACKGROUND OF THE INVENTION
Robotic devices can enable today to assist the movement of a user in many areas such as medical, security national On thus knows exoskeletons worn by the user and participating in his movements for him make it easier to perform tasks such as wearing a object, pushing or pulling of an object A
exoskeleton makes it possible to artificially increase the forces of the user and thus limit his fatigue.
So that an exoskeleton can effectively help a user, it is however necessary to know the external forces applied as the mass of objects transported. It is indeed advisable to be able to order the exoskeleton so that it can automatically compensate these efforts as accurately and quickly as possible.
Usually, we estimate the external forces via additional indicator elements arranged in environment or on the exoskeleton (camera, technology radio-identification better known by the English acronym RFID, strain gauges _) but also linked to user (myoelectric technology, user interface user-environment _).
This therefore obliges the user to wear directly from sensors and / or interacts with a interface making it less ergonomic to wear the exoskeleton.
Furthermore, the use of the exoskeleton is confined to less complex applications such as example a static application in the factory allowing a

2 easier placement of markers, typically to provide information automatically the mass and physical properties of an object to wear.
OBJECT OF THE INVENTION
An aim of the invention is to provide a method assistance of at least one movement of the user which is more ergonomic for the user.
An object of the invention is also to provide a corresponding device.
BRIEF DESCRIPTION OF THE INVENTION
In order to achieve this goal, we propose a method of assisting at least one movement of a user through a robotic device, the method of operating the device robotics at least in the following two modes:
- Default mode: force control of the robotic device via a computer of the device robotic, - Measurement mode: slaving of the trajectory of the robotic device, via the computer, and recording in the calculator of at least one force exerted on the robotic device by an external environment audit robotic device, the process being defined for, when the computer recognizes a movement performed by the robotic device, switch from the default mode to measurement mode to record at least a force exerted on the robotic device by a environment outside the robotic device, then return to default mode with the help of said effort registered to implement force control of said default mode.
In this way, the invention makes it possible to switch regularly between a slaving in trajectory and a force control, trajectory control

3 thus temporarily allowing to be able to inform the calculator on the external force (s) applied to the robotic device and the force control to assist the user in his movement.
This therefore makes it possible to participate in the proper functioning of the robotic device.
The invention thus makes it possible to limit the number indicator elements worn directly by the user or with which it must interact.
The inventors were thus able to develop a robotic device operating without a worn sensor directly by the user, in particular without sensor biometric, and also without interface user / environment through which the user traditionally identified the level of assistance that the robotic device was to bring him.
The invention therefore makes movement assistance more ergonomic for the user.
In addition, the invention can be used for relatively complex applications. It can thus be intended to use the invention in an environment unknown.
According to a particular embodiment, the ECU performs automatic learning of at least minus one movement performed by the robotic device.
According to a particular embodiment, automatic learning is carried out during mode by fault.
Thus, switching between the two modes enslavement thus becomes more and more frequent to as the computer learns the possible movements of the robotic device making the robotic device of more and more efficient over time.
According to a particular embodiment, machine learning is performed during a phase

4 preliminary dedicated.
According to a particular embodiment, trajectory control is a following slavings in position.
According to a particular embodiment, the trajectory control is a control in speed.
According to a particular embodiment, the robotic device is an exoskeleton.
The invention also relates to a robotic device.
comprising a computer configured to control in force the robotic device, the computer being arranged for, during a recognition of a movement made by the robotic device, enslave the device in trajectory robotic while registering at least one exerted effort by an external environment on the robotic device, in order, during a return to control in effort, to using this recording to ensure the enslavement in effort.
Other characteristics and advantages of the invention will emerge on reading the following description of a particular non-limiting embodiment of invention.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention will be better understood in the light of description which follows with reference to the attached figures joined among which:
- Figure 1 is a schematic view of a device robotic according to a particular embodiment of invention, - Figure 2 schematically illustrates the passage between the two operating modes of the device illustrated in figure 1, - Figure 3a schematically illustrates the device shown in Figure 1 when operating in par mode default, and - Figure 3b schematically illustrates the device

5 shown in figure 1 when operating in the measure.
DETAILED DESCRIPTION OF THE INVENTION
With reference to FIG. 1, the method according to a mode particular embodiment is implemented here in a exoskeleton 1.
This application is of course not limiting and the process can be implemented in any type of robotic device, such as a coding robot handling, an aircraft guidance system or any other means of transport ... The device may also be a sighting system with tracking of a target. In supplement or replacement of follow-up by treatment images, said system will thus be able to benefit from the applied by the user to a guide member (as a handle, a joystick for monitor the target. For example, as soon as a trajectory of the organ of guidance can be recognized by the computer, said calculator switches to calculation mode to determine the forces exerted on the guide member. The device robotics (or robot) can thus be shaped to provide an additional effort to the user in order to compensate for the effort required to interact with objects and / or perform tiring muscular efforts and / or painful.
The exoskeleton 1 comprises here in a manner known in self :
- a mechanical structure 2 with one or more degrees of freedom that is carried by the user, - one or more actuators 3 associated with the mechanical structure 2 to allow the displacement of the mechanical structure 2 WO 2019/02526

6 PCT / EP2018 / 070225 according to the degree (s) of freedom of the mechanical structure 2, and - a computer 4 controlling the actuators 3.
The actuators 3 are for example motors.
According to a particular embodiment, the exoskeleton 1 further comprises at least one trajectory 5 (to estimate the position and / or the speed of displacement of the mechanical structure 2). The sensor trajectory 5 is for example a position sensor angular. The trajectory sensor 5 is here arranged at the level of the pivot connection between two articulated elements of mechanical structure 2 in order to measure the position relative angular of one of the elements with respect to the other.
Preferably, the exoskeleton 1 also comprises at least a force sensor 6 (to estimate one or more forces (torque and / or force) applied to the mechanical structure 2).
Said sensor 6 is for example a torque sensor. He is located here between the actuator 3 allowing the pivoting between the two articulated elements of the mechanical structure 2 and the mechanical structure 2 itself.
With reference to Figures 2, 3a and 3b, the exoskeleton 1 is configured to be able to operate in two modes different, a default mode and a measurement mode.
In the default mode, computer 4 provides servo-control of the exoskeleton 1. By effort we hear that enslavement is just as well a servo in force than a servo couple.
Preferably, in this default mode, the computer 4 simultaneously performs automatic learning of the or movements made by exoskeleton 1, which are therefore linked to those of the user.
Remember that machine learning (for machine learning in English) is a method allowing

7 to evolve through a systematic process, in place of the use of classical algorithms too limited for complete the intended task. This type of learning is good known to those skilled in the art and will therefore not be detailed here.
Typically, the computer 4 records in a base data from movement models performed by the exoskeleton 1 during the enslavement in effort (by example when the torque applied to the user by exoskeleton 1 is zero).
Regarding the measurement mode, calculator 4 slaves exoskeleton 1 in trajectory.
Through trajectory we hear that the enslavement is everything as well a series of servos in position (angular or not) than a speed control (by example by following a speed curve). We prefer here enslave in a series of servos in position the exoskeleton 1.
It should be noted that we are talking here about slavings in position due to the fact that one seeks to make the exoskeleton 1 reproduce a movement, movement which is therefore defined by a succession of positions.
At the same time, in the measuring mode, the calculator 4 records the force (s) exoskeleton 1 by the environment outside the exoskeleton 1 for reinject this data later in force control when switching to par mode fault.
Note that in this measurement mode, calculator 4 predicts the movement of the user and slaves in correspondingly position the exoskeleton 1. This minimizes the direct impact of the user on the measurement of the forces exerted on exoskeleton 1 by the environment outside the exoskeleton 1.

8 Preferably, in this measurement mode, we record in computer 4 at least one force exerted on exoskeleton 1 by an external environment audit exoskeleton 1 if we know the direction of said force and at least two efforts to determine this direction if we don't know her. In reality, said at least two forces result from the action of the same object on exoskeleton 1 but measured at different times:
the exoskeleton 1 having moved between these two moments, the effort will have evolved in the meantime, hence the notion of two efforts . By comparing the effort measurements made to two different times, computer 4 can thus estimate the result and the direction of the action exerted by the object on the exoskeleton 1.
The operation of exoskeleton 1 will be present described.
In the initial state, exoskeleton 1 is in its default mode. Exoskeleton 1 thus assists the user in the performance of his tasks by his enslavement in effort. Simultaneously, calculator 4 automatically learns the movements performed by exoskeleton 1 and saves them in its database.
When computer 4 determines that a movement performed by exoskeleton 1 corresponds to a movement that computer 4 has already learned and saved, the computer 4 switches to measurement mode.
Typically the signals supplied by the trajectory 5 are processed by computer 4: when of the signals corresponds to a signal known to the computer 4, said computer 4 switches the exoskeleton 1 into the mode of measure.
The computer 4 also continuously extrapolates the following the received signal, thanks to one or more models of movements contained in its database, to estimate the movement that exoskeleton 1 is supposed to perform. A

9 from this presumed movement of exoskeleton 1, the computer 4 provides control in the trajectory of the exoskeleton 1.
When switching to the measurement mode it is therefore exoskeleton 1 which causes movement and no longer the user. From then on exoskeleton 1 alone supports the or the forces applied to the exoskeleton 1 by the external environment. As indicated above, we considers that the efforts directly exerted by user on exoskeleton 1 in measurement mode are negligible (thanks to machine learning the trajectory control in fact compensates for the effort user's muscle).
The data then supplied by the force sensor 6 are thus significant of the effort (s) that exoskeleton 1 must apply alone to the environment to be able to follow the trajectory imposed by the computer 4. From the data supplied by the sensor force 6, the computer 4 thus deduces the force (s) exerted by the external environment on the exoskeleton 1 and that exoskeleton 1 must compensate.
For example, if the user is carrying a heavy object without making any muscular effort, his arm tends to unfold, but in reality exoskeleton 1 will retain its arm in the measuring mode thanks to the action of the actuators 3. By a micro-twist effect at the level of the sensor 6, the torque induced by the object is thus measured door.
Once this determination has been made, the computer 4 causes the switch to the default mode. The effort of the user is again added to that of the exoskeleton 1. The computer 4 then uses the determination of the stress (s) exerted by the environment exterior on exoskeleton 1 determined during operation in the measuring mode to ensure the control in force and compensate, if necessary, the forces exerted by the external environment on the exoskeleton 1. More precisely, if the measured effort during measurement mode is different from estimated one 5 in the default mode, the forces exerted are increased by actuator (s) 3 (by adjusting their powers) to get closer to the force measured in the mode of measure. When switching on exoskeleton 1, we considers for the default mode that no effort is

10 exerted by the external environment on the exoskeleton 1.
Exoskeleton 1 then operates in its mode by fault until a known movement is detected by the calculator 4.
An example of how it works is carrying a object.
When the computer 4 recognizes the movement of the user signifying that he is about to enter an object or grab an object, the calculator 4 switches to measuring mode.
The computer 4 then measures the force that must be exercise alone exoskeleton 1 to carry the object (since thanks to machine learning, automatic control trajectory compensates for the user's muscular effort) and deduces the mass of the object.
We then switch back to default mode and the computer 4 uses the measured mass of the object to ensure adequate force servo-control.
Of course this is a particular example of functioning. More generally, calculator 4 may cause switching to the measuring mode for other movements such as pushing or to shoot an object. Computer 4 will then seek to measure the force and direction of this force exerted by the object on the exoskeleton 1.

11 The computer 4 thus regularly causes switching from one operating mode to another ensuring assistance with the user's movements in a precise, dynamic and adapted over time. According to a mode of particular embodiment, the computer 4 is configured to switch to measurement mode at each recognition by the computer 4 of a movement learned by said calculator 4.
Advantageously, the operation in the mode measurement only lasts a short time, typically between 0.3 and 3.5 seconds, and preferably between 0.5 and 3 seconds, just enough time to determine the efforts exteriors exercised. This makes it possible to make the operation in measurement mode little, if any, perceptible for the user. This makes the exoskeleton 1 more ergonomic for the user. In addition, this makes it possible to no need to predict the user's movement during too important a phase.
It is noted that the exoskeleton 1 thus described does not include only two sensors, one for movement and the other for the efforts.
In addition, the exoskeleton 1 thus described functions without user interface / environment through which the user identifies the level of assistance that must be bring exoskeleton 1. In addition, exoskeleton 1 as well described works without sensor carried directly by the user, in particular without a biometric sensor.
The exoskeleton 1 thus described therefore turns out to be particularly ergonomic.
Besides, exoskeleton 1 can function even if the user progresses in an unfamiliar environment.
For example the mass of the objects to be carried does not need to be predetermined so that exoskeleton 1 can function.

12 Of course, the invention is not limited to the method of embodiment described and variations can be made without depart from the scope of the invention as defined by the claims.
In particular, although here in the default mode the ECU performs automatic learning of at least minus one movement made by the robotic device in during operation of the robotic device, in variant or in addition, the computer can perform automatic learning of at least one movement performed by the robotic device during a dedicated phase learning (the robotic device then being only used for learning the computer and not to assist a user in his tasks). In variant or in addition, the computer can also integrate a database including a list predetermined pattern of movements performed by the robotic device in force control.
In addition, the device may include or be associated with a greater number of force sensors and / or trajectories than what has been indicated. We will seek however to limit the number of indicator elements for questions of cost, feasibility, ergonomics, maintenance Although here everything is on board by the device robotics, some of the elements (such as a power supply of the device _) can be deported from the device.
Although the device is able to operate without = 30 sensors worn directly by the user, the user can wear sensors for example for questions redundancy in the event of device failure.

Claims (8)

13 1. Method of assisting at least one movement of a user via robotic device (1), the robotic device comprising at least one sensor trajectory and at least one force sensor, the method enabling the robotic device to operate at the less in the following two modes:
default mode: force control of the device robotics via a computer (4) of the robotic device, measurement mode: slaving of the trajectory of the robotic device, via the computer, and recording in the calculator of at least one force exerted on the robotic device by an external environment audit robotic device, the process being defined for, during a reconnaissance, from data provided by at least the trajectory sensor, by the computer of a movement made by the robotic device, already learned and saved by the computer, switch between mode defaults to measuring mode to record at least one force exerted on the robotic device by the environment external to the robotic device using data supplied by at least the force sensor, then return to default mode with the help of said effort registered to implement force control of said default mode. 2. The method of claim 1, wherein the computer (4) performs automatic learning of at least minus one movement made by the robotic device and in which machine learning is performed during default mode. 3. The method of claim 1, wherein the computer (4) performs automatic learning of at least minus one movement made by the robotic device and in which machine learning is performed during a dedicated preliminary phase.
Date Received / Date Received 2021-04-19 4. The method of claim 1, wherein the computer (4) performs automatic learning of at least minus one movement made by the robotic device, machine learning being performed during a phase preliminary dedicated then being carried out during the mode by fault. 5. Method according to any one of claims 1 to 4, in which the trajectory control is a succession of slavings in position. 6. Method according to any one of claims 1 to 4, in which the trajectory control is a speed control. 7. Method according to any one of claims 1 to 6, in which the robotic device is a exoskeleton. 8. Robotic device (1) comprising a computer (4), at least one trajectory sensor and at least one force sensor, the computer being configured to control the robotic device, the computer being arranged for, upon recognition of a movement performed by the robotic device already learned and recorded by the computer, enslave the robotic device while registering at least one effort exerted by an external environment on the device robotics, in order, during a return to the enslavement in effort, to use this recording to ensure enslavement in effort.
Date Received / Date Received 2021-04-19